KR101476018B1 - Buffer disk in flashcopy cascade - Google Patents

Abstract

A method of operating a copy function includes initiating a new flashcopy function from a source volume to a target volume, detecting that the target volume of the new flashcopy function is a source volume for an existing flashcopy function, Detecting that the target volume of the function has a secondary volume, and creating a buffer flashcopy function from the target volume of the new flashcopy function to a new target volume.

Description

BUFFER DISK IN FLASHCOPY CASCADE}

The present invention relates to a method and system for operating a copy function.

Storage of data in large organizations is of fundamental importance for both data reliability and data recovery in the event of hardware failure. A storage area network (SAN) is an architecture used when very large amounts of data need to be stored in a reliable and stable manner. This technology allows remote computer storage devices to be connected to servers, such as to connect disk arrays to servers, to make them appear to be locally attached to an operating system Networks to be created can be created. It is common in these networks to include a large amount of redundancy, both in terms of hardware connections between data storage and individual components.

There are various ways to create data redundancy. For example, a function, such as a flashcopy function, allows an administrator to set a point-in-time, full volume copies of data at some point (point-in-time) , And the copies are available for read or write access immediately. Flashcopy can be used with standard backup tools available in an environment that creates backup copies on magnetic tapes. The FlashCopy function creates a copy of the source volume on the target volume. This copy is referred to as a point-in-time copy, as described above. When the flashcopy operation is initiated, a relationship is created between the source volume and the target volume. This relationship is a "mapping" between the target volume and the source volume. This mapping allows the point-in-time copy of the source volume to be copied to the associated target volume. The relationship between these volume pairs exists from when the flashcopy operation is started until the storage unit copies all of the data of the source volume to the target volume, otherwise this relationship is lost.

When data is physically copied, a background process copies tracks from the source volume to the target volume. The time it takes to complete a background copy depends on various criteria, such as the amount of data to be copied, the number of background copy processes in progress, and other activities currently in progress. The data to be copied does not need to be actually instantly copied when the FlashCopy function is active, but it needs to be copied before an update that results in an overwrite of the old data on the source volume. So, when there are data changes on the source volume, the original data is copied to the target volume, which is copied before overwriting on the source volume.

Thus, FlashCopy is a feature supported on various storage devices, which makes it possible for a user or an automated process to make nearly instantaneous copies of the entire logical volume of data. A copy (copy) of the source disc is created on the target disc. Copies are immediately available for both read and write access. A common feature of implementations, such as flashcopy, is the ability to reverse the copy. In other words, it populates the source disk of the flashcopy map with the contents of the target disk. It is also possible to use FlashCopy in cascaded implementations, where the target disk becomes the source disk for further FlashCopy later, or vice versa.

It is desirable to provide a data structure that defines primary and secondary " fdisks " to avoid missing such cascaded storage volumes and flashcopy functions. The FEF disk defines a storage volume associated therewith and contains a logical component (a) that includes an index that provides a link to the associated maps that define the up and down directions of the flashcopy functions within the cascade logical component. When a flashcopy function is created between a source volume and a target volume, primary FEF disks are created for each storage volume, unless primary FEF disks already exist for the target disk, If it already exists, the existing FEF disk for the target volume is converted to a secondary FEF disk and a new primary FEF disk is created. The advantage of using the data structures defined by the FEF disks is that they do not miss the IO read and write accesses to other storage volumes within the existing multiple cascades, An FEP disk can be used to direct data reads to the correct location.

The limitation of flashcopy cascade is that it is required to limit the number of concurrent restore operations to limit the number of clean operations required for a given write operation. (A, B, C, and D are disks in the graph, arrows are flashcopy maps, and (A, B) is a flashcopy mapping from disk A to disk B , The cascade has maps of (A, B), (B, C) and (C, D). In this cascade of disks and flashcopy functions, a write to disk A results in a split write to disk B, which is required to keep the image on disk B, Clean reads of disks B and C and clean writes to disk D following disk C In this way, a single lead to the top disk in a cascade can go further on the storage volumes to further the down the cascade to a large number of clean operations. Can be imported. It is therefore an object of the present invention to improve such prior art.

A first object of the present invention is to provide a method of operating a copy function, the method comprising initiating a new flashcopy function from a source volume to a target volume, Detecting a source volume for a flashcopy function, detecting that the target volume of the existing flashcopy function has a secondary volume, and detecting a new target from the target volume of the new flashcopy function Creating a < / RTI > buffer with a buffer flashcopy function (further the down the cascade).

A second object of the present invention is to provide a system for operating a copy function, the system comprising a plurality of storage volumes and a storage volume controller coupled thereto, the storage controller having a new flashcopy function from a source volume to a target volume Detecting that the target volume of the new flashcopy function is the source volume for an existing flashcopy function and detecting that the target volume of the existing flashcopy function has a secondary volume, And to create a buffer flashcopy function from the target volume of the function to a new target volume.

A third object of the present invention is to provide a computer program product on a computer readable medium for operating a copy function, said product initiating a new flashcopy function from a source volume to a target volume, Detecting that the target volume is the source volume for an existing flashcopy function, detecting that the target volume of the existing flashcopy function has a secondary volume, and detecting a new target volume of the new flashcopy function And instructions for creating a buffer flashcopy function.

Thanks to the invention it is possible to provide buffered flashcopy maps that enable unbounded flashcopy recovery operations. The system and method of the present invention describe a procedure for removing restrictions in prior art implementations of FlashCopy cascades. In the prior art implementations of the cascades, one light will cause large clean operations to descend cascade, which will slow the completion of the original write action at the top of the cascade (slow down). The present invention introduces the concept of buffered FlashCopy. In other words, a new space efficient FlashCopy is created and started when the FlashCopy is started on the source of another active map with the target having a secondary volume, Thereby preventing spreading to the whole.

These methods add an extra step to the beginning of the FlashCopy map, where the target volume is the source volume of an existing active FlashCopy map. This step queries whether the target of the flashcopy map to be started, X, is the source of the active flash map, 1, the target of the map 1, Y, has a secondary EF disk, and if the answer is yes (if so) to create a buffered flash copy from X to the new space efficient vdisk X '.

Considering an example of the prior art of the present invention and a novel configuration of the present invention, when (B, C) starts, the cascades B↔C and C↔D are generated (resulting in) The map's target, C, is part of map C↔D, but D does not have a secondary volume. Now, when (A, B) starts, a buffer flashcopy function (B, B ') is created and started because target B is part of B → C and C has a secondary volume. This new buffer flashcopy function creates cascaded A↔B, B↔B'↔C and C↔D. Once this is created, a new write to disk A will bring the clean lead to disk B and clean light to disk B '. No matter how large the FlashCopy graph is, a single light will bring only one clean operation. The map (B, B ') will be in a permanent cleaning mode. This means that all data on disks B 'or C will be cleaned in the background

The buffer flash copy map will exist for at least the map A? B and B? C and C? D lifetimes. If A → B or C↔D is interrupted or completed, map B↔B 'cleans itself from the cascade and removes it. If B ↔ C is interrupted or completed, the map B ↔ B 'can be interrupted immediately. This means that the cleaning required to maintain the cascades is independent of the number of interlocked cascades. Of course, it is also possible to extend this idea to perform additional clean writes (beyond that described in the above example) to reduce the number of interlocked per-cascade buffer FlashCopy maps. This is a practical consideration.

BRIEF DESCRIPTION OF THE DRAWINGS Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which: Fig.
Figure 1 is a schematic diagram of a pair of storage disks,
Figure 2 is a schematic of a flashcopy cascade,
Figure 3 is a schematic of an extended FlashCopy cascade,
Figure 4 is a schematic diagram of an extended FlashCopy cascade with a data light,
5 is a flow chart of a method of operating a copy function.

Figure 1 illustrates the concept of a flash copy using a storage controller 8 and two storage disks 10 and 12. The disks 10 and 12 may constitute part of a larger array of disks and may typically constitute part of a corporate storage solution. The disks 10 and 12 may, for example, be part of a storage solution associated with a commercial web site. A flashcopy command is issued from the storage volume controller 8 to the disk 10 defining the source disk 10 (Vdisk 1), if the backup is to be made with the contents of vdiskl at any time. And the target disk 12 (V-disk 2), which is the target of the flash copy. The flashcopy command produces a point-in-time copy of the image of the specific vdisk that is source disk 10.

In the example of FIG. 1, the source disk 10 of the first flashcopy command is V-disk 1, and the target disk 12 is V-disk 2. The flashcopy command starts a flashcopy process that creates a map 14 from the source disk 10 to the target disk 12. [ This map is labeled as map (1) in the figure. At this specific point in time, the image of Vdisk 1 is now available on Vdisk 2. This creates a backup of the data on the V disk 1 and also allows the execution of tests and other administration tasks on the data of V disk 1 as an attendant danger of the loss of original data, , Since the original data is stored on the original source disc.

When a flash copy is made, it creates a link between the two disks 10 and 12, as defined by the map 14. The data may also be copied in the background, subject to the additional requirements below. That is, any access to Vdisk 2 (as the target disk 12) immediately causes the images of the relevant portions of the Vdisk 1 to be copied, and also to bring changes to the image stored by the disk 10 Any access to V disk 1 causes the unaltered data to be copied to the target disk 12 immediately before the change takes place. In this way, although Vdisk 2 is physically copied under the circumstances described above, Vdisk 2 stores a point-in-time copy of Vdisk 1 for an external user.

The storage volume, which is the target volume of the flashcopy function, can also be the source volume of the further flashcopy function, thus creating a cascade of storage volumes. Figure 2 shows a cascade example of four storage volumes 10, 12, 16 and 18, which are connected to respective FlashCopy maps 14. Each map 14 defines a flashcopy function from a source volume to a target volume. Disk B provides backup of disk A, disk C provides backup of disk B, and disk D provides backup of disk C. The flashcopy functions 14 connecting the other storage volumes may have been started at different times, which creates another point-in-time copy of the images stored by each of the storage volumes.

(Where A, B, C and D are the disks in the cascade and the arrows are flashcopy maps, as shown in FIG. 2), so that A, B, ) Indicates a flashcopy mapping from disk A to disk B), the cascade has maps A, B, (B, C) and (C, D). In the prior art implementation of such a cascade, any new data write to disk A would result in a split write to disk B for each flashcopy function , Which is required to keep the image on disk B. This writing to disk B adds clean reads to disks B and C followed by clean writes to disk D followed by a write to disk C. < RTI ID = 0.0 > further. In this manner, a single write to the first storage volume 10 in the cascade may result in large-scale clean operations throughout the cascade.

Therefore, the limitations of such prior art FlashCopy cascades are that it is necessary to limit the number of concurrent restore operations in order to limit the number of clean operations required for a given write operation will be. Since the writes to disk A are the normal running of the services supported by storage volume A, it is important from a business perspective that these writes are completed as quickly as possible. In the cascade of FIG. 2, the write to disk A can not be completed until all of the dependent reads and writes have been completed, as described above, If either fails, the whole transaction needs to be backed up.

FIG. 3 shows how the configuration of FIG. 2 is extended to improve the problem of the delay in completing the initial write to Vdisk A. FIG. The storage volume controller 8 adds an extra step to the start of the flashcopy map, where the target volume is already the source volume of the active flash map. This step effectively queries the target volume of the new map being started to determine whether the target volume is the source volume of the active FlashCopy map and whether the target volume of the active map has a secondary FEP And if so, the storage volume controller will create a buffer flash copy from the target volume of the original flash copy to the new space efficient V disk. Secondary is as defined above. Vdisk has two images that it can provide. These are referred to as FEF disks. A primary FEF disk is an image provided to all host systems. It is the data returned for read operations. A secondary FEF disk is an image used by other FlashCopy maps that require data held on other V disks to provide their own images.

Referring to the example of FIG. 2 described above, when (B, C) is started, and as a result the cascades B? C and C? D are generated, in the new scheme, The target, C, is part of map C↔D, but there is no problem because D does not have a secondary volume at this point. However, when (A, B) starts, a buffer flashcopy function (B, B ') is created because target B has a portion of B → C and C has a secondary volume. This new buffer flashcopy function uses the new storage volume 20 to create a cascade that includes A↔B, B↔B'↔C and C↔D. Once this is created, the light on disk A will bring a clean lead to disk B and a clean light to disk B '. No matter how large the FlashCopy graph is, a single write will only bring a single clean operation. The map (B, B ') will be in a permanent cleaning mode. This means that all data on disk B 'or disk C will be cleaned in the background.

The buffer FlashCopy map (B, B ') will be present for at least the duration of the map A → B and B → C and C → D. If A → B or C↔D is interrupted or completed, map B↔B 'cleans itself from the cascade and removes it. If B↔C is interrupted or completed, the map B↔B 'can be immediately interrupted. This means that the cleaning required to maintain the cascade is independent of the number of interlocked cascades. A new target disc 20 that is the target of the buffer flashcopy function will effectively create a break in the original cascade and will accept the changes requested from disc B resulting from the original write to disc A. [ This light can then be completed and a clean operation from B 'to C and below the cascade can be performed.

Figure 4 shows how the write to disk A is handled once the buffer flashcopy function is set up. The presence of the new target disk 20, V disk B ', creates a boundary for IO for disk A. A new light on disk A brings the clean lead to disk B and the clean light to disk B '. No further actions are required at this point for discs C or D below the cascade. (Nevertheless) the original IO for disk A can be completed, which means that when compared to prior art cascades of a number of disks in series, the time required to complete the original IO to disk A Lt; RTI ID = 0.0 > length.

The presence of the buffer flashcopy functions and the new target storage volume 20 means that any restrictions on the order of starting the flashcopy maps have been removed. Storage volume B 'acts as a brake in the cascade, and once the original IO is completed, the data on volume B' can be cleaned onto disk C as a normal background process. Storage volume B 'is a temporary store for data written from disk B, and the data provided to storage volume B' is not persistent after it has been cleaned to disk C.

In Cascade, the lower volumes below function normally, and so are the maps (14) between these storage volumes. In the examples of FIGS. 3 and 4, discs C and D, which are lower down in the cascade, do not recognize the presence of a new target disc B 'inserted in the cascade and also do not recognize the presence of a buffer flashcopy function. Work on these disks C and D proceeds normally and a clean operation of the data from the new target disk B 'to disk C is processed as a normal write of the data for that disk C, which triggers the flashcopy function, If the specific data on C has not yet been copied, it will cause the write to take place on disk D.

In the set-up stage, a flow chart summarizing the process of operating the copy function is shown in FIG. The method of operating the copy function, performed by the storage volume controller 8, comprises a first step S1, wherein step S1 includes initiating a new flashcopy function from a source volume to a target volume . In the example of Fig. 4, the source volume is vdisk A and the target volume is vdisk B, and the new flashcopy function that is created becomes flashcopy function 14 from vdisk A to vdisk B. This new FlashCopy function may be created by the administrator or automatically by software.

The second step S2 in the method is a step of detecting that the target volume (Vdisk B) of the new flashcopy function is also the source volume for the existing flashcopy function. In the context of the example of FIG. 4, the existing flashcopy function is a function of mapping from Vdisk B to Vdisk C. Therefore, in this example, Vdisk B is the target of the new flashcopy function, and at the same time becomes the source of the existing flashcopy function. The storage volume controller 8 may perform this detection step, which is based on the details maintained by the controller 8 in relation to existing mappings of flashcopy functions from source volumes to target volumes .

The next step in the process is step S3, which is the step of detecting that the target volume (Vdisk C) of the existing FlashCopy function (B → C) has a secondary volume (Vdisk D in this case). This can also be done by the storage volume controller 8, using flashcopy functions and existing data on their sources and targets. Finally, a finishing step is provided in step S4, which creates a buffer FlashCopy function (B - > B ') from the target volume (Vdisk B) of the new FlashCopy function to the new target volume . In this way, the brakes are introduced into the cascade of the V-disk B ', and the IO for the original disk A is now bounded.

Claims (15)

A method of operating a copy function, the method comprising:
Initiating a new flashcopy function from the source volume to the target volume,
Detecting that the target volume of the new flashcopy function is a source volume for an existing flashcopy function,
Detecting that the target volume of the existing flashcopy function has a secondary volume, and
Creating a buffer flashcopy function from a target volume of the new FlashCopy function to a new target volume
Way.
2. The method of claim 1, further comprising: receiving a data write for the source volume; performing a corresponding data read of the target volume; and performing a data write of the data lead from the target volume to the new target volume Further comprising
Way.
3. The method of claim 2, further comprising: performing a corresponding data read of the source volume, performing a data write to the target volume, and completing an original data write for the source volume
Way.
4. The method of claim 3, further comprising performing a data write on a target volume of the existing flashcopy function of data written for the new target volume
Way. The method of claim 1, further comprising: detecting that either the new FlashCopy function or the existing FlashCopy function has been interrupted or completed; cleaning the new target volume; and removing the new target volume from the copy function Further comprising
Way.
A system for operating a copy function comprising a plurality of storage volumes and a storage volume controller coupled thereto, the storage volume controller comprising:
Initiating a new flashcopy function from the source volume to the target volume,
Detecting that the target volume of the new FlashCopy function is a source volume for an existing FlashCopy function,
The target volume of the existing FlashCopy function detects that it has a secondary volume, and
And to create a buffer flashcopy function from a target volume of the new flashcopy function to a new target volume.
system.
7. The method of claim 6, wherein the storage volume controller is configured to receive a data write for the source volume, perform a corresponding data read of the target volume, RTI ID = 0.0 >
system.
8. The method of claim 7, wherein the storage volume controller is further configured to perform a corresponding data read of the source volume, perform a data write to the target volume, and complete an original data write for the source volume
system.
9. The system of claim 8, wherein the storage volume controller is further configured to perform a data write on the target volume of an existing flashcopy function of the data written to the new target volume
system.
delete A computer-readable medium comprising:
The computer readable medium comprising a computer program for operating a copy function, the computer program comprising a computer program for causing a computer to perform all steps of the method recited in any one of claims 1 to 5, ≪ / RTI > delete delete delete delete

Images